UncTrack

The official implementation of the TIP 2025 paper UncTrack: Reliable Visual Object Tracking with Uncertainty-Aware Prototype Memory Network

📢 Announcement: This paper has been accepted by IEEE Transactions on Image Processing (TIP)! 🎉

[Models and Raw results] (Google Driver)
[Models and Raw results] (Baidu Driver: 4409)

Install the environment

Use the Anaconda

conda create -n unctrack python=3.7
conda activate unctrack
bash install.sh

Data Preparation

Data should be prepared in the following format:

-- lasot
   |-- airplane
   |-- basketball
   |-- bear
    ...
-- got10k
   |-- test
   |-- train
   |-- val
-- coco
   |-- annotations
   |-- train2017
-- trackingnet
   |-- TRAIN_0
       -- anno
       -- frames
   |-- TRAIN_1
    ...
   |-- TRAIN_11
   |-- TEST
-- nat2021
   |-- test
       -- anno
       -- data_seq
       -- list.txt
-- uav123
   |-- anno
       -- UAV123
   |-- data_seq
       -- UAV123
   |-- frame
-- otb100
   |-- Basketball
    ...

Set project paths

Run the following command to set paths for this project

python tracking/create_default_local_file.py --workspace_dir . --data_dir ./data --save_dir .

After running this command, you can also modify paths by editing these two files

lib/train/admin/local.py  # paths about training
lib/test/evaluation/local.py  # paths about testing

Train UncTrack

Training with multiple GPUs using DDP. More details of other training settings can be found at tracking/train_unctrack.sh for different backbone respectively.

bash tracking/train_unctrack.sh

Test and evaluate UncTrack on benchmarks

• LaSOT/GOT10k-test/TrackingNet/OTB100/UAV123/NAT2021. More details of test settings can be found at tracking/test_unctrack.sh

bash tracking/test_unctrack.sh

• VOT2020
  Before evaluating "UncTrack+AR" on VOT2020, please install some extra packages following external/AR/README.md. Also, the VOT toolkit is required to evaluate our tracker. To download and install VOT toolkit, you can follow this tutorial. For convenience, you can use our example workspaces of VOT toolkit under external/vot20/ by setting trackers.ini.

cd external/vot20/<workspace_dir>
vot evaluate --workspace . UncTrackPython
# generating analysis results
vot analysis --workspace .

Compute FLOPs/Params and test speed

bash tracking/profile_model.sh

Model Zoo and raw results

The trained models and the raw tracking results are provided in the [Models and Raw results] (Google Driver) or [Models and Raw results] (Baidu Driver: 4409).

Contact

Yang Guo: guoyang4409@gmail.com

Siyuan Yao: yaosiyuan04@gmail.com

Acknowledgments

• Thanks for MixFormer Library, which helps us to quickly implement our ideas.

✏️ Citation

If you think this project is helpful, please feel free to leave a star⭐️ and cite our paper:

@misc{unctrack,
      title={UncTrack: Reliable Visual Object Tracking with Uncertainty-Aware Prototype Memory Network}, 
      author={Siyuan Yao and Yang Guo and Yanyang Yan and Wenqi Ren and Xiaochun Cao},
      year={2025},
      eprint={2503.12888},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2503.12888}, 
}